Twitch Chat AI Monitor

A real-time sentiment and toxicity analysis engine for Twitch, developed as a final-year engineering project at the Faculté Polydisciplinaire de Khouribga (FPK). git add README.md

🚀 Project Overview

This project addresses the challenges of high-velocity live stream moderation by providing an AI-driven, real-time analysis of chat sentiment. The system categorizes chat activity into Hype, Neutral, and Toxic sentiment patterns, enabling content creators to reduce moderator burnout and maintain safe, inclusive community interactions.

🛠 Technical Methodology

The system architecture focuses on speed and accuracy:

1. Ingestion Layer: Persistent TCP socket connections (IRC Protocol) for real-time data streaming.
2. Preprocessing Firewall: A custom heuristic TWITCH_DICT layer that normalizes gaming-specific slang and emotes (e.g., "Pog", "LULW") into standard linguistic vectors before AI processing.
3. Inference Engine: A fine-tuned DistilBERT transformer model, optimized to provide classification results in under 200ms.
4. Telemetry & UI: A Streamlit-powered dashboard that visualizes sentiment spikes and provides actionable chat telemetry logs.

⚙️ Tech Stack

• Language: Python 3.11.9
• Deep Learning: PyTorch, Hugging Face Transformers
• Visualization: Streamlit
• Data Processing: Pandas, Regex
• Protocol: Twitch IRC API

👥 Development Team

• BAHRI Ilyass
• EL HICHAMI Oussama

Supervised by: Dr. BAKKOURI Ibtissam, Faculty of Sciences and Techniques of Khouribga (FPK).

---

🛠 Installation Guide

Prerequisites

• Python Version: 3.11.9

Setup Instructions

1. Clone the repository:

git clone [https://github.com/ilyassbaa/Twitch-Chat-AI-Monitor.git](https://github.com/ilyassbaa/Twitch-Chat-AI-Monitor.git)
cd Twitch-Chat-AI-Monitor

2. Create a virtual environment:

python -m venv env
# On Windows:
env\Scripts\activate
# On macOS/Linux:
source env/bin/activate

3. Install dependencies:

pip install -r requirements.txt

4. Run the application:

streamlit run app.py

---

📊 Dashboard Overview

📜 License

This project is licensed under the MIT License. See the LICENSE file for details.

🏷 Keywords

#Twitch #AI #MachineLearning #NLP #SentimentAnalysis #ToxicityDetection #StreamerTools #SoftwareEngineering #Python #PyTorch #HuggingFace #DistilBERT #Streamlit #DataScience #TechInternship #FPK #RealTimeSystems #ChatModeration #Innovation